grnn神經網路

A. GRNN（廣義回歸神經網路）和BP神經網路

一般用SOM網路 推薦你一本書 神經網路43案例 里頭有數字識別這個

B. matlab grnn預測神經網路怎麼用

輪台歌奉送封大夫出師西征(岑參)[4]

C. elman神經網路能夠解決的問題,還有其他什麼網路能夠更好的解決

還可以使用GRNN神經網路，效果非常好，並且訓練速度非常快。廣義回歸神經網路GRNN：徑向基神經內元容和線性神經元可以建立廣義回歸神經網路，它是徑RBF網路的一種變化形式，經常用於函數逼近。在某些方面比RBF網路更具優勢。
在MATLAB中，直接使用net=newgrnn(P,T,spread)就能以非常快的速度設計出一個GRNN網路，其進行訓練及預測時，效果非常好，不會比elman神經網路差。擴展常數SPREAD不能太小，才能使部分徑向基神經元能夠對輸入向量所覆蓋的區間產生相應，但也不能太大，否則計算困難。可以通過試湊來獲得最佳擴展常數。

D. 請教RBF神經網路高手。用matlab設計newgrnn廣義回歸神經網路，進行訓練、模擬、擬合。畫出預測圖和誤差圖

這個擬合了也沒有多大的意義。
一、數據太少。二、發病率和時間存不存在因果關系還是個疑問

E. 幫忙翻譯一下啊！謝謝了！基於GRNN神經網路的中厚板軋機寬展預測模型

Plate Plate Mill is the main rolling equipment, shoulders rolling steel pressure vessels, Offshore Steel, ships and other high-quality steel-plate proction tasks. In order to improve the proction process of rolling automation level and proction efficiency, need for the rolling process of plate deformation accurate prediction and control. Spread change prediction model is rolling width control technology set the width of the core functions of calculation, its accuracy will have a direct impact on the width of the finished proct control effect. Only by correctly calculated and the estimated spread of size, can be rolled out in line with the requirements of the procts. According to the same size and principles, spread by the size of the volume can be calculated directly for the blank size, So research process of rolling wide changes in the size of the development is of great significance. Is necessary considering the plate mill modern proction methods of operation and modern art, Forecast to achieve automatic control system SL process of deformation spread. Spread a theoretical analysis and development of wide deformation refers to the rolling workpiece width along the direction of deformation. Rolling in different conditions, the blank in the process of rolling spread along the height of the cross-sectional analysis shown in Figure 1. Spread △ Says generally following components : sliding spread △ Yue. The figure △-wide outreach and drum-wide outreach △ B3 [11 o spread is a complex deformation of the deformation process. in the process of rolling wide impact on the development of many factors. Spread with a series of rolling factors have complicated relations : △ B = f (H, h, l, B, D, ψa, Δh. ε, f, t, m, Pδ, V, ε) (I) where : a ψ, H, h for the SL system before and after the deformation zone thickness; I, B, D deformation zone for the length, width and roll diameter; deformation of the cross-sectional shape; Δ h, ε, to pass, Rection; . F, t, m coefficient of friction, rolling temperature, the chemical composition of metal; P δ for the metal mechanical properties; V, ε roller linear velocity and deformation rate. From the above analysis shows that the sheet rolling process, not only with the width of the thickness rection has increased, sheet width but also by the process of rolling many factors, these factors between the relations are very complicated, difficult to traditional mathematical models to achieve changes in the spread forecast precision. If considering various parameters on the workpiece width of change, using a mathematical model expression is extremely complicated, can not be directly applied to engineering practice. For 42 oo ~ L machine width, the mathematical model for the information have also introced, Researchers at home and abroad to establish the spread mathematical model has a very wide range, but mainly on the basis of the measured data of the scene, Spread consider the influence of changes in several key factors, ignore other factors, the use of regression analysis of the mathematical model, This is bound to cause certain error; and the spread of existing model are limited in their scope of application. Neural network is highly nonlinear processing capability, operation and high accuracy on-line real-time response characteristics, In describing the spread on the deformation has obvious advantages. Therefore, the deformation of the spread of automatic detection and forecast process, the use of artificial neural network, introction GRNN (GRNN) model algorithm to achieve the width deformation forecast, Rolling to the actual proction process spread deformation control provide precise control basis.

F. 在RBF神經網路預測中，如果樣本數據較少時，效果怎麼樣和GRNN神經網路比較

效果都不好,樣本少最好是用統計學加一些機械學習的思想自己多做嘗試,設計回特徵給出一個答固定的預測模型,效果根神經網路不在一個檔次上,人的智商遠遠比電腦高,神經網路勉強算大數據的技術,前提就是要有大量冗餘的數據,大到疲於用統計學方法處理.

G. 急需一個GRNN神經網路的C的源代碼，有誰能幫我寫一下。實在是看不懂。。。

抱歉！這個實在不容易，沒有

H. 通用回歸神經網路是啥類似BP神經網路嘛

一般的數據擬合，傳統的lsqcurvefit和lsqnonlin，如果較新的方法就很多了，比如神經網路，小樣本的一般是GRNN和灰色神經網路，大樣本下更多選擇了，BP、SVM等等，還可以有遺傳演算法等等

I. matlab神經網路43個案例分析 百度雲搜索

共有43章，內容涵蓋常見的神經網路(BP、RBF、SOM、Hopfield、Elman、LVQ、Kohonen、GRNN、NARX等)以及相關智能演算法(SVM、決策樹、隨機森林、極限學習專機等)。同時，部分章節也涉及了常見的優化演算法(遺傳演算法、蟻群演算法等)與神經網路的結合問題。此外，本書還介紹了MATLABR2012b中神經網路工具箱的新增功能與特性，如神經網路並行計算、定製神經網路、屬神經網路高效編程等。